A 2³ full factorial design, replicated twice, was used to study the effects of the homogenization method (high-pressure valve homogenization or microfluidization), modified sunflower lecithin type (DL or HL), and emulsion type (monolayer or bilayer) on each variable studied. The experimental design and the sample codes are shown in Table 1.

The fatty acid composition of chia seed oil was analyzed by gas chromatography (GC) according to International Union of Pure and Applied Chemistry (IUPAC) 2.302. Fatty acid methyl esters (FAMEs) were prepared using boron trifluoride-methanol (cat# 373-57-9, Alfa Chemistry, Holbrook, NY, USA) reagent following the IUPAC 2.301 method [33]. The peroxide value and the free fatty acid content were also determined according to American Oil Chemists Society (AOCS) recommended practices [34, 35].

Phosphorus-31 nuclear magnetic resonance (³¹P NMR) spectroscopy was employed to determine the PL composition of both modified sunflower lecithin (DL and HL), according to Diehl [36]. Briefly, 100 mg of each sample was diluted in 1 mL of deuterated chloroform, 1 mL of methanol, and 1 mL of 0.2 mol/L Ch-ethylenediaminetetraacetic acid (Cs-EDTA). The mixture was shaken for 15 min, afterwards, the organic layer was separated and analyzed using a Bruker Avance spectrometer (Bruker Avance 600 MHz automatic spectrometer, UK) and triphenyl phosphate as the internal standard.

Initially, about 500 mL of stock solutions of DL and HL were prepared by dispersing each in 35.46 mmol/L acetic acid (cas# 64-19-7, Cicarelli, Santa Fe, Argentina)/64.54 mmol/L sodium acetate (cas# 127-09-3, Merck, Darmstadt, Germany) buffer solution (pH 5.0). Similarly, 80 mL of a stock biopolymer solution was obtained by dissolving Ch in the same buffer solution and stirring for around 8 h. Afterward, the emulsifying and biopolymer stock solutions were refrigerated at 4°C ± 0.5°C overnight to ensure complete hydration.

The optimal amounts of DL and HL, as well as the suitable pH conditions for facilitating the electrostatic deposition of Ch onto the modified lecithin-stabilized droplets during the formation of double-layered nanoemulsions, while minimizing the residual excess of the emulsifying agent in the continuous phase, were determined based on the results of preliminary experiments (data not shown).The shows the scheme of mono and bilayer chia oil-in-water (O/W) nanoemulsions preparation is presented in Figure 1. In this way, about 250 mL of each O/W nanoemulsion containing 5% w/w of chia seed oil and stabilized with a simple (0.5% w/w of DL or HL) and double layer (0.5% w/w of DL or HL, and 0.3% w/w of Ch) were produced by applying a two-step homogenization process and the LBL deposition technique in the latter case. In the first step, a pre-homogenization was performed using a laboratory scale disperser (Ultraturrax T-25, IKA-Labortechnik, GmbH & Co., Staufen, Germany) for 3 min at 13,000 rpm. The coarse emulsion obtained by ultraturrax was subsequently treated by either a microfluidizer (LM10, Microfluidics, Westwood, CA, USA) or a HPH (Panda 2 K, GEA, NiroSoavi, Parma, Italy) at 1,000 bar for three passes (Figure 1). During the homogenization process, the temperature of the nanoemulsions was maintained below 20°C by employing an ice-water bath. Food-grade preservatives potassium sorbate (1,000 ppm, cas# 24634-61-5, Cicarelli, Argentina) and nisine (12 ppm, cas# 1414-45-5, El Maestro Quesero, Argentina) were added to the nanoemulsions obtained. Finally, all systems were stored for 30 days at 4°C ± 0.5°C in darkness.

Chia mono and bilayer nanoemulsions were analyzed by confocal laser scanning microscopy using a confocal microscope (cat# FV1000, Olympus, Shinjuku, Tokyo, Japan). Samples (about 1 mL) were stained by directly adding Nile Red (cat# 7385-67-3, Sigma Aldrich, St. Louis, MO) and allowing them to rest in the dark. Excitation and emission wavelengths of 488 nm and 518 nm, respectively, were employed. Micrographs were captured at 20× magnifications. The analysis of the images was carried out using the OlyVIA software (3.7 version, Olympus, Japan).

To extract chia oil from the nanoemulsions, 3 mL of each system was mixed with 15 mL of an isopropanol/isooctane (1:3 v/v) mixture. The mixture was then centrifuged at 5,000 g for 2 min. Following centrifugation, the organic phase was collected, and the solvent was subsequently removed using a rotary evaporator (Rotavapor R 110, Buchi) under a pressure range of 95–100 bar at 50°C. Finally, any remaining solvent and surrounding oxygen were eliminated using a stream of nitrogen (N₂). After oil extraction, α-linolenic acid content was determined by proton nuclear magnetic resonance (¹H NMR) according to Julio et al. [8]. Briefly, 100 mg of each nanoemulsion was dissolved in 1 mL of the mixture deuterated chloroform (DCl₃)/methanol-d4 (MeOD)/Cs-EDTA-deuterium oxide (D₂O) 0.2 mol/L (1:1:1), ultrasonicated for 30 min, and shaken for at approximately 1 h. The mixture was centrifuged, and the organic layer was analyzed using a nuclear magnetic resonance (NMR) spectrometer Avance III 500 MHz (Bruker, Karlsruhe, Germany), magnetic flux density 11.7 with Tesla BFOPLUS SmartProbe (5 mm PABBO BB) at 25°C. The calibration was performed on the tetramethylsilane (TMS) signal. Data were analyzed using the software Bruker TopSpin 3.2, standard operation procedure SAA-GMR045-03. ¹H NMR spectroscopic measurements were performed immediately after their preparation and after 30 days of storage in triplicate.

The zeta-potential serves as an electrokinetic potential that quantifies the extent of electric charge on the surface of emulsion droplets, which is a vital determinant of their stability [37]. This parameter was assessed for various chia oil nanoemulsions at pH 5.0, confirming the complete electrostatic deposition of Ch onto the oil droplets stabilized by modified sunflower lecithin using the LBL technique. The surface charge of droplets in the different chia nanoemulsions is presented in Table 3. Both the addition of Ch and the homogenization method had a significant impact (P ≤ 0.05) on the surface charge of chia oil droplets. As expected, before the Ch deposition, one-layer nanoemulsions exhibited a negative surface charge, which can be attributed to the anionic PLs from DL or HL at pH 5.0. Conversely, after the addition of Ch, the oil droplets of double layer nanoemulsions exhibited a positive charge which may be due to the amino groups (-NH₃⁺; pK_a < 6.3) present in molecules of this cationic biopolymer suggesting thus the electrostatic deposition of Ch onto the entire surface of the droplets of monolayer systems [2]. Furthermore, the droplets coated by double-layer exhibited a relatively high positive charge (+18.37 mV to +34.26 mV), creating a robust electrostatic repulsion between them, effectively preventing aggregation. In contrast, the droplets in single-layer systems carried a relatively low negative charge at pH 5.0 (–5.95 mV to –12.14 mV), which could be insufficient to avoid droplet aggregation.

Additionally, the choice of homogenization method also had a significant impact (P ≤ 0.05) on the zeta-potential of the oil droplets. Notably, the absolute value of the zeta-potential in emulsions produced through high-pressure valve homogenization was higher (P ≤ 0.05) than those generated through microfluidization. This divergence may be attributed to the distinct conformational arrangement of the modified lecithin and Ch at the droplet interfaces, resulting in different geometries of the devices. Nevertheless, despite the droplets produced by microfluidization having a lower charge than those from high-pressure valve homogenization, it would be sufficient to provide stability against creaming and flocculation processes.

In this study, the formulations and homogenization conditions applied produced emulsions with nanoscale droplet sizes since the D [3, 2] values ranged from 110–229 nm, as presented in Table 3.

The initial mean diameter was influenced (P ≤ 0.05) by the three factors investigated to varying degrees, with the order of impact as follows: homogenization method > emulsion type > lecithin type. When the emulsifier content and homogenization pressure were kept constant, microfluidization produced significantly smaller droplet sizes (P ≤ 0.05) than high-pressure valve homogenization (Table 3). Lee and Norton [15] also observed similar results, reporting that the microfluidizer produced smaller droplets and a narrower distribution of particle sizes compared to the HPH at 100 MPa and 150 MPa. This contrast is due to the limited range of shear forces generated by the microfluidizer compared to the broader spectrum of forces produced by the HPH in food-grade emulsions. Furthermore, these findings align with prior research by Ciron et al. [38] examining the impact of conventional homogenization and microfluidization on particle size in milk and other dairy-based emulsions. Their study revealed that microfluidization is conducive to producing smaller milk droplets than conventional homogenization. It was noted that as pressure increased, emulsion droplet sizes gradually decreased, and microfluidizer-generated emulsions exhibited significantly smaller droplet sizes in comparison to those produced via HPH under similar homogenization conditions. Other studies have also shown that microfluidization leads to emulsions with smaller droplets, attributed to the ease of emulsifier adsorption on the surface during microfluidization [39]. This fact can be attributed to the distinctive geometry of the microfluidizer, where oil droplets in the interaction chamber experience substantial shearing forces, approximately 15 times higher than those during valve homogenization. The extended capillary tube in the chamber ensures effective emulsifier adsorption [39].

When examining the influence of emulsion type on particle size, it was observed that the mean particle diameter of nanoemulsions produced via high-pressure valve homogenization significantly decreased (P ≤ 0.05) with the addition of Ch to form the interfacial double layer. This phenomenon can be attributed to the electrostatic deposition of this biopolymer on anionic droplets coated with modified sunflower lecithin. As a result, the oil droplets became highly positively charged, potentially impeding recoalescence during the homogenization process through electrostatic repulsion forces. These findings align with those of Wang et al. [40], who investigated bilayer emulsions containing rice bran protein hydrolysate and Ch. They reported that the application of adequate Ch can result in the formation of a complete covering film on the surface of monolayer droplets. This technique has the potential to effectively reduce particle size by preventing emulsion aggregation and flocculation, thanks to the positively charged nature of Ch.

The type of sunflower lecithin also exerted a significant influence on droplet size, albeit to a lesser extent. Specifically, when high-pressure valve homogenization was employed, the use of HL for forming nanoemulsions resulted in smaller droplet sizes (P ≤ 0.05) compared to those obtained with DL. Previous research by Cabezas et al. [41] has demonstrated that two types of HL sunflower lecithin produced droplet mean diameters significantly smaller than those formed with DL sunflower lecithin, especially at levels of 2% w/w. This trend can be attributed, as previously discussed, to the HLB value associated with emulsifiers. The higher concentration of hydrophilic PLs, such as lysophospholipids, in HL increases the empirical HLB value, thereby enhancing its performance as an O/W emulsifying agent.

Moreover, the specific arrangement of different PLs at the oil-water interface significantly influences emulsion formation and stability. PC typically forms stable monolayers or bilayers, while LPC and LPE form hexagonal widespread clusters that facilitate O/W emulsion stabilization. Conversely, PE gives rise to a reversed hexagonal phase, which is more challenging to arrange at the interface [42]. Therefore, the higher content of PE in DL could explain its reduced efficacy as an emulsifying agent compared to HL. In line with D [3, 2] results, the span values were significantly affected (P ≤ 0.05) by the three examined factors (Table 3). Also, all factor interactions were statistically significant, with the type of emulsion exerting the most substantial influence. Bilayer systems exhibited lower span values than monolayer systems, notably evident when employing a HPH in system preparation. Consequently, these findings imply potential recoalescence events during the homogenization of monolayer systems, potentially leading to broader particle distribution curves and the emergence of larger droplet populations.

The PSD curves corresponding to the freshly prepared oil nanoemulsions are shown in Figure 2. As can be seen, systems formed by microfluidization resulted in the PSD curves shifting toward smaller particle sizes compared to high-pressure valve homogenization. This behavior was more evident in systems stabilized with DL (Figure 2A) than in HL (Figure 2B). Besides, nanoemulsions obtained by microfluidization presented unimodal (double-layer nanoemulsions) and trimodal (one-layer nanoemulsions) PSD curves, while those formed by high-pressure valve homogenization were bimodal (one-layer nanoemulsions) and trimodal (double-layer nanoemulsions). Moreover, upon the introduction of Ch and the development of an interfacial double layer, a distinct reduction in the number of peaks associated with larger droplets in the distribution curves becomes evident. This observation also suggests a decrease in coalescence events during the homogenization process.

The flow behavior of the chia nanoemulsions was assessed at a controlled temperature of 25°C ± 0.5°C. The n and K parameters were obtained from the rheological data fitted to the power-law model (Table 3). According to this, all one-layer emulsions behaved as Newtonian fluids (n = 1), while bilayer ones resulted in pseudoplastic fluids since n < 1. When considering the K coefficient, it was primarily influenced by the type of emulsions, with a relatively minor impact from the homogenization method used. Specifically, emulsions with a double-layer structure, prepared using microfluidization technology, exhibited significantly higher values (P ≤ 0.05) for this parameter. Additionally, for a meaningful comparison of the viscosity among the various nanoemulsions, the apparent viscosity at a shear rate of 100 s^–1 (η₁₀₀) is presented in Figure 3. This parameter holds significance in food science and processing as it offers essential insights into the behavior of a food product during mastication, processing, and quality control [43]. It significantly influences sensory attributes and the efficiency of manufacturing processes. The η₁₀₀ values of the nanoemulsions, in general, were relatively low, indicating that the droplets in the emulsions did not flocculate, and the dispersion of the entire emulsion system was good. Similar to the K coefficient, the η₁₀₀ values were significantly influenced (P ≤ 0.05) by the addition Ch and the homogenization method. The Ch addition to form the double layer in nanoemulsions substantially increased η₁₀₀ in both the systems with DL and HL. This increase can be attributed to the structure formed by a three-dimensional network of interconnected Ch chains [44]. Furthermore, the apparent viscosity of double-layered nanoemulsions obtained by microfluidization exhibited higher η₁₀₀ values (P ≤ 0.05). These results can be linked to the smaller droplet size in these systems, with increased droplet concentration and interfacial surface area, consequently enhancing the strength of the Ch network.

The physical stability of nanoemulsions was determined by analyzing their BS profiles over 30 days. These profiles are depicted as the %BS in relation to the tube length. Upon immediate preparation, the mono-layer systems exhibited an initial %BS (average of initial BS values along the entire sample tube, denoted as BS₀) of approximately 76–78%, while the bilayer ones had an initial %BS of 82–84% (Figure 4).

Immediately after the emulsion preparation, it can be considered a homogeneous PSD and thus associate their BS₀ with their mean droplet diameter. According to this, when the droplet size is smaller than the incident wavelength (λ = 0.8 μm), it is expected that the %BS increases [45]. Thus, it would explain the lower values of BS₀ found in the former, which presented larger droplets.

In order to spot signs of destabilization processes such as creaming or flocculation, the analysis of their BS profiles in two distinct zones within the sample tube, zone I (bottom of the sample tube) and zone II (top of the sample tube) was conducted (Figure 4). The bilayer nanoemulsions demonstrated remarkable physical stability, as evidenced by their unchanged BS profiles. Throughout the entire storage period, there were negligible variations in the average %BS values at the bottom and top of the sample tube. The high physical stability of these systems may be accounted for by different reasons. One of them could be the high surface charge of their oil droplets which would cause enough electrostatic repulsion between droplets and thus prevent the nanoemulsion coalescence. Another plausible explanation could be the smaller droplet size and higher apparent viscosity of bilayer nanoemulsions compared to monolayer systems. This combination may reduce droplet mobility and, as a result, hinder their upward movement, as predicted by Stokes’s law. In accordance with findings from previous studies, the Ch incorporation to create the double interfacial layer enveloping the oil droplets resulted in an elevation of the viscosity of the continuous phase. This viscosity increase contributed to the enhancement of creaming stability by retarding the diffusion of droplets and flocs [46].

Conversely, in the case of monolayer systems, a continuous creaming process was observed, starting as early as day 5 and continuing over time, resulting in the formation of a cream phase. The non-uniform distribution of oil droplets along the sample tube was confirmed by a decrease in the %BS in zone I and a simultaneous increase of this parameter in zone II, signifying destabilization due to creaming (Figure 4). Consequently, a gradual clarification process in the lower section of the sample tube (zone I) and the formation of a cream layer in zone II were noted. This phenomenon is attributed to the migration of oil droplets towards the upper zone, driven by the lower oil density in comparison to the aqueous phase. These observations may be linked to the lower apparent viscosity and zeta-potential recorded for the monolayer systems compared to their bilayer counterparts.